A rapid evaluation method for the mechanical properties of automotive engine hoods based on machine learning

Abstract

The present application discloses a method for rapidly evaluating the mechanical properties of an automotive engine hood based on machine learning. The method includes: performing multiple trainings and tests based on an original data set and multiple initial algorithm models, and extracting initial feature vectors of the multiple initial algorithm models and corresponding prediction data sets; calculating and generating an intermediate training data set according to the initial feature vectors and the prediction data sets; calculating a termination center point set corresponding to the initial center point set when a preset condition function converges according to the initial center point set of the original data set and the preset condition function; generating a training data set according to the termination center point set and the intermediate training data set; calculating a hybrid intelligent model corresponding to the minimum value of the objective function value according to the training data set, wherein the hybrid intelligent model is used for rapidly evaluating the mechanical properties of the automotive engine hood. Through the technical solution in the present application, the problems of long R & D cycle, low prediction accuracy and low stability of the engine hood are solved.

Description

Technical Field

[0001] This application relates to the technical field of automobiles. Specifically, it relates to a method for quickly evaluating the mechanical properties of an automobile engine hood based on machine learning. Background Technique

[0002] With the rapid development of the global automobile industry, the automobile design cycle is continuously shortened. Traditional automobile component design methods are difficult to meet the rapid iteration of the market. The rapid design method of components can provide solutions to this problem. As an important part of the vehicle body structure, the performance of the automobile engine hood directly affects the use performance of the whole vehicle. In particular, the stiffness and modal of the engine hood need to meet the requirements of the whole vehicle performance. Therefore, it is of great value and significance to study the mechanical properties of the engine hood.

[0003] In the existing design process of the engine hood structure, first, CAD forward design is required, and then through CAE simulation, its torsional stiffness and first-order mode are verified to determine whether its design parameters meet the requirements of the whole vehicle performance. The design process is not only complex and time-consuming, but also, as the design parameters in the engine hood structure increase, the time consumed will increase exponentially.

[0004] In addition, when one or some design parameters do not meet the requirements of the whole vehicle performance, it often depends on the rich engineering experience of engineers for adjustment, and the uncertainty in the process of adjusting the design parameters is still inevitable, which prolongs the R & D cycle of the engine hood. Summary of the Invention

[0005] The purpose of this application is to apply the prediction results of multiple machine learning algorithm models in a fused manner to the evaluation of the mechanical properties of the engine hood, so as to solve the problems of the long R & D cycle of the engine hood and the low prediction accuracy and stability.

[0006] The technical solution of this application is: to provide a method for quickly evaluating the mechanical properties of an automobile engine hood based on machine learning. The method includes: Step 1, based on the original data set and multiple initial algorithm models, conduct multiple trainings and tests, and extract the initial feature vectors of the multiple initial algorithm models and the corresponding prediction data sets; Step 2, calculate and generate an intermediate training data set according to the initial feature vectors and the prediction data sets; Step 3, calculate the termination center point set corresponding to the initial center point set when the preset condition function converges according to the initial center point set of the original data set and the preset condition function; Step 4, generate a training data set according to the termination center point set and the intermediate training data set; Step 5, calculate the hybrid intelligent model corresponding to the minimum value of the objective function value according to the training data set, where the hybrid intelligent model is used for quickly evaluating the mechanical properties of the automobile engine hood.

[0007] In any of the above technical solutions, further, in step 2, according to the initial feature vector and the prediction data set, calculate and generate an intermediate training data set, specifically including: step 21, calculate the prediction mean value of the prediction data set corresponding to each training and test, and the maximum value in the prediction mean value; step 22, based on the initial feature vector, the maximum value in the initial feature vector, the prediction data set, and the maximum value in the prediction mean value, calculate and generate the intermediate training data set.

[0008] In any of the above technical solutions, further, in step 2, the intermediate training data set D new(n×k) The corresponding calculation formula is:

[0009]

[0010] s M =Max{R ij}

[0011]

[0012] In the formula, D new(n×k) is the intermediate training data set, S M is the maximum value in the initial feature vector {R ij}, is the prediction mean value of the prediction data set {Pred ij}, V M is the maximum value of the prediction mean value in the prediction data set , R ij is the element in the i-th row and j-th column of the feature vector {R ij}.

[0013] In any of the above technical solutions, further, in step 3, it specifically includes: step 31, using a clustering algorithm to determine the initial center point set of the original data set; step 32, according to the initial center point set, determine the category to which each training data in the original data set belongs; step 33, according to the initial center point set and the category to which it belongs, in an iterative manner, calculate the termination center point set corresponding to the initial center point set when the preset condition function converges.

[0014] In any of the above technical solutions, further, in step 33, the termination center point set {E} new is composed of multiple center points μ e , the termination center point set {E} new The calculation formula of the center point μ e is:

[0015]

[0016] In the formula, m is the training data x mwith the number m = 1, 2, …, M, c m is the category to which it belongs, and J(c, μ) is a preset condition function.

[0017] In any of the above technical solutions, further, before step 5, it further includes: converting the training data set X in terms of the expression mode, and sequentially extracting the termination center point set {E} new and the intermediate training data set D new(n×k) the elements at the corresponding positions in are formed into an array form, and the training data set X is converted to:

[0018] The calculation formula of the hybrid intelligent model is:

[0019]

[0020] In the formula, is the predicted value of the hybrid intelligent model, J(ω) is the objective function, ω is a preset constant, i′ is the row number, r is the total number of rows, x i′ is the element in the termination center point set {E} new in the i′-th row, y i′ is the element in the i′-th row of the data in the intermediate training data set D new(n×k) and z is the CAE software simulation value corresponding to the input data.

[0021] In any of the above technical solutions, further, the initial feature vector includes at least a first-order modal feature vector and a torsional stiffness feature vector.

[0022] The beneficial effects of this application are:

[0023] In the technical solution of this application, multiple machine learning algorithm models are applied to the mechanical property evaluation of the engine hood. Based on the initial feature vectors of multiple initial algorithm models, a new training data set is formed by using a data fusion algorithm, and a hybrid intelligent model is formed by restricting the objective function and the new training data set. This model constructs the mapping relationship between the design parameters and the mechanical properties of the engine hood, and can quickly and accurately predict the first-order modal performance and torsional stiffness performance of the engine hood.

[0024] In a preferred implementation of the present application, taking the automotive engine hood as the research object, making full use of the initial feature vectors and the corresponding prediction data sets, and setting the corresponding preset condition functions, the mapping relationship between the mechanical properties of the engine hood and its key design parameters is mined to determine the optimal training data set, and then the final hybrid intelligent model is obtained. The advantages of various machine learning algorithms are fully utilized, which is applicable to the mechanical property evaluation of various engine hoods, and has the characteristics of intelligence, high prediction accuracy and good stability. It realizes the rapid intelligent evaluation of the stiffness, strength and mode of the engine hood, and can help engineers quickly determine the best design parameters, providing guidance and reference for the rapid verification of automotive engine hood engineering simulation. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] The above and / or additional aspects of the present application will become apparent and be readily understood from the following description of the embodiments in conjunction with the accompanying drawings, in which:

[0026] Figure 1 is a schematic flow chart of a method for rapid evaluation of the mechanical properties of an automotive engine hood based on machine learning according to an embodiment of the present application;

[0027] Figure 2 is a schematic diagram of design parameters and their positions on the engine hood according to an embodiment of the present application;

[0028] Figure 3 is a schematic diagram of the prediction results of training the initial algorithm model according to an embodiment of the present application;

[0029] Figure 4 is a block diagram of calculating the hybrid intelligent model according to an embodiment of the present application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0030] In order to more clearly understand the above objects, features and advantages of the present application, the present application will be further described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be noted that, without conflict, the embodiments of the present application and the features in the embodiments can be combined with each other.

[0031] In the following description, many specific details are set forth in order to fully understand the present application. However, the present application may be implemented in other ways different from those described herein. Therefore, the protection scope of the present application is not limited by the specific embodiments disclosed below.

[0032] Under normal circumstances, the engine hood structure has more than 10 key parameters. According to statistics, during the design process of the engine hood structure, it takes about 20 seconds to modify a design parameter, about 3 minutes to modify a group of design parameters, and about 2 minutes to complete the performance simulation of a group of design parameters. Overall, it takes about 5 minutes to modify a group of design parameters and verify their performance, which often relies on the rich engineering experience of engineers. Therefore, in order to improve the design efficiency of the engine hood, this embodiment can achieve intelligent evaluation of the mechanical properties of various engine hoods, thereby shortening its development cycle.

[0033] As Figure 1 shown, this embodiment provides a method for quickly evaluating the mechanical properties of an automotive engine hood based on machine learning. By integrating the advantages of other machine learning algorithms and intelligently matching each advantage, a hybrid learning model is finally constructed. The method specifically includes:

[0034] Step 1, based on the original data set {D n×m} and multiple initial algorithm models, conduct multiple trainings and tests, and extract multiple initial feature vectors {R ij} and the corresponding prediction data set {Pred ij}, where the initial feature vectors at least include the first-order modal feature vector {R ij} M and the torsional stiffness feature vector {R ij} T ;

[0035] Specifically, select any algorithm model, such as: machine learning algorithms such as Extreme Gradient Boosting (XGBoost), Deep Neural Network (DNN), Random Forest (RF), Support Vector Regression (SVR), and K-Nearest Neighbor Algorithm (KNN), as the initial algorithm model, and conduct multiple trainings and tests on the original data set {D n×m}.

[0036] Set the original training data set {D n×m} as an N×M matrix, which is determined by each design parameter in the engine hood design process, and its corresponding performance data set is {P n×k}, which is an N×K matrix.

[0037] As Figure 2 shown, in this embodiment, the design parameters included in the original training data set {D n×m} are shown in Table 1.

[0038] Table 1

[0039]

[0040] In this embodiment, the simulation conditions for the simulation experiment are set as follows: CPU, AMD Ryzen-74800H (2.90 GHz); RAM, 16.0 GB, Windows 10.

[0041] In this embodiment, co-simulation is performed using Abaqus and Isight software. Since the contour dimensions of the engine hood must meet the assembly relationship of the whole vehicle and cannot be changed, 11 key dimensional parameters of the internal structure of the engine hood are selected as the dependent variables affecting the mechanical properties of the engine hood.

[0042] In addition, those skilled in the art can understand that the torsional stiffness reflects the anti-deformation ability of automotive-related components, and the improvement of torsional stiffness can reduce the deformation degree of the whole vehicle frame when loaded.

[0043] The engine hood in the front compartment position is a key factor affecting the vehicle NVH performance, and the value of the first-order mode of the engine hood is the key to the design. Therefore, how to effectively increase the first-order mode value is the key to engineering design.

[0044] Therefore, in this example, the torsional stiffness and the first-order mode of the engine hood are used as research indicators in the form of data tags, which can study the influence of the design parameters of the engine hood on them. Optimizing and improving this indicator can improve the mechanical and NVH performance of the whole vehicle.

[0045] In this embodiment, combinations of 11 design parameters are obtained by the Latin hypercube sampling method and their corresponding tag values are obtained through simulation, creating a prediction data set of the mechanical properties of the engine hood with 6959 sample data, as the original training data set {D n×m}, that is, {D 6959×11}. Correspondingly, the performance data set {P n×k} is {P n6959×2}.

[0046] After multiple trainings and tests on each initial algorithm model, the results of each training and test are extracted, and the feature vectors of the trained models are extracted as the initial feature vectors {R ij}, and the corresponding original training data set is used as the prediction data set {Pred ij} corresponding to the initial feature vector {R ij}.

[0047] This embodiment involves W machine learning algorithms, such as: Extreme Gradient Boosting (XGBoost), Deep Neural Network (DNN), Random Forest (RF), Support Vector Regression (SVR), and K-Nearest Neighbor (KNN), etc. The machine learning algorithms are denoted as W i , i = 1, 2,..., W.

[0048] Set each algorithm model to perform 10 times of training and testing from the original training dataset {D n×m} to the performance dataset {P n×k}, and for each time, the corresponding algorithm performance metrics R 2 , mean absolute error MAE, and root mean square error RMSE can be obtained. That is, these three parameters constitute the feature vector output by the algorithm model, namely the initial feature vector {R ij}, where the feature vector of the first-order mode is denoted as {R ij} M , and the feature vector of torsional stiffness is denoted as {R ij} T , i represents the model of the i-th machine learning algorithm, and j represents the j-th training and testing, j = 1, 2,..., 10.

[0049] It should be noted that the specific implementation process of each machine learning algorithm in this embodiment is not limited.

[0050] As Figure 3 shown, taking the above five machine learning algorithms as an example, in the j-th training and testing, the predicted performance corresponding to the first-order mode in the feature vectors of each model is shown in Table 2, and the predicted performance corresponding to torsional stiffness is shown in Table 3.

[0051] Table 2

[0052]

[0053]

[0054] Table 3

[0055] Prediction model <![CDATA[R 2 > MAE RMSE KNN 0.8998 0.05350 0.06693 SVR 0.9333 0.43550 0.04569 RF 0.9549 0.03543 0.03889 DNN 0.9643 0.02259 0.02853 XGBoost 0.9756 0.01117 0.01802

[0056] For the first-order mode performance, the feature vector {R ij} M corresponding to the first-order mode in the j-th training and testing of each model is:

[0057] {R 1j} M = {0.8816, 4.00624, 5.11298}

[0058] {R 2j} M = {0.9214, 3.27109, 4.16501}

[0059] {R 3j} M = {0.9625, 1.55750, 2.46483}

[0060] {R4j} M = {0.9717, 1.38608, 2.21335}

[0061] {R 5j} M = {0.9821, 0.82268, 1.83339}

[0062] For the torsional stiffness performance, the eigenvector {R ij} T corresponding to the j-th training and testing of each model for torsional stiffness is:

[0063] {R 1j} T = {0.8998, 0.05350, 0.06693}

[0064] {R 2j} T = {0.9333, 0.43550, 0.04569}

[0065] {R 3j} T = {0.9549, 0.03543, 0.03889}

[0066] {R 4j} T = {0.9643, 0.02259, 0.02853}

[0067] {R 5j} T = {0.9756, 0.01117, 0.01802}

[0068] It should be noted that since the initial eigenvector in this embodiment includes at least the first-order modal eigenvector {R ij} M and the torsional stiffness eigenvector {R ij} T , therefore, when generating the subsequent training dataset X, the corresponding training dataset X can be generated respectively according to the first-order modal eigenvector {R ij} M and the torsional stiffness eigenvector {R ij} T so that in the hybrid intelligent model, when training the algorithm model for the first-order mode, the first-order modal eigenvector {R ij} M is used; when training the algorithm model for torsional stiffness, the torsional stiffness eigenvector {R ij} T is used. It can also be based on the first-order modal eigenvector {Rij} M and the torsional stiffness eigenvector {R ij} T Generate a training data set X for training in the hybrid intelligent model, and the specific training process is not limited in this embodiment.

[0069] Step 2, according to the initial eigenvector {R ij} and the prediction data set {Pred ij}, calculate and generate the intermediate training data set D new(n×k) ;

[0070] In this embodiment, after each machine learning algorithm completes any training and testing, it will perform data prediction on the original training data set {D n×m} to obtain the output result obtained by the algorithm for the original training data set {D n×m}, and then obtain a corresponding set of predicted values Pred ij ), where the predicted value Pred ij is a set of matrix, N and K are the number of rows and columns of the performance data set {P n×k}.

[0071] After 10 times of training and testing are completed, based on 10 sets of predicted values Pred ij}, the prediction data set {Pred ij} will be formed. The prediction data set {Pred ij} is an N×K matrix, and the data in this matrix are the predicted values of the first-order mode and torsional stiffness of the engine hood by each trained machine learning algorithm.

[0072] Specifically, this step 2 includes:

[0073] Step 21, calculate the prediction mean ij of the prediction data set {Pred } corresponding to each training and testing, and the maximum value V in the prediction mean M ;

[0074] Step 22, based on the initial eigenvector {R ij}, the maximum value S ij in the initial eigenvector {R M}, the prediction data set {Pred ij}, and the maximum value V in the prediction mean M , calculate and generate the intermediate training data set D new(n×k) .

[0075] Specifically, based on the prediction data set {Predij} and the eigenvector {R ij}, generate the intermediate training dataset D new(n×k) , and the corresponding calculation formula is:

[0076]

[0077] S M = Max{R ij}

[0078]

[0079] In the formula, D new(n×k) is the intermediate training dataset, S M is the maximum value in the initial eigenvector {R ij}, is the predicted mean of the prediction dataset {Pred ij}, V M is the maximum value of the predicted mean in the prediction dataset , and R ij is the element in the i-th row and j-th column of the eigenvector {R ij}.

[0080] Step 3: According to the initial center point set {E} of the original dataset {D n×m} and the preset condition function, calculate the termination center point set {E} corresponding to the initial center point set {E} when the preset condition function converges new ;

[0081] Specifically, this step 3 includes:

[0082] Step 31: Use the clustering algorithm to determine the initial center point set {E} of the original dataset {D n×m};

[0083] Step 32: According to the initial center point set {E}, determine the category c n×m to which each training data x m in the original dataset {D m} belongs;

[0084] Specifically, determine the initial center point set {E} of the original training dataset {D n×m}, where each center point is denoted as {μ 1 , μ 2 ,..., μ e , …, μ E}, μ e is the center point of the initial center point set {E}, and e is the e-th center point in the initial center point set.

[0085] Set the original training dataset {Dn×m} = {x 1 , x 2 ,..., x m ,…, x M}, where x m is the m-th training data, and x m ∈ R. Using the idea of the clustering algorithm, the Euclidean distance between a certain point and the surrounding central points is used as a reference for clustering. In order to calculate the category c n×m to which all the training data in the original training data set {D m} belong, where c

[0086]

[0087] Step 33: According to the initial central point set {E} and the category c m to which they belong, an iterative method is used to calculate the termination central point set {E} new corresponding to the initial central point set {E} when the preset condition function converges.

[0088] Specifically, according to the calculated category c m to which they belong, each central point μ e in the central point set {E} is updated, and the corresponding calculation formula is:

[0089]

[0090] Repeat the above process until the condition function converges to determine the termination central point set {E} new ,

[0091] where the calculation formula of the preset condition function J(c, μ) is:

[0092]

[0093] In the formula, is the central point position of the category where the m-th data is located, c is the category to which the m-th data belongs, and μ is the central position.

[0094] Therefore, in the termination central point set {E} new , the calculation formula of the central point μ e is:

[0095]

[0096] In the formula, m is the number of the training data x m , m = 1, 2,..., M, c m is the category to which it belongs,

[0097] Step 4, according to the set of termination center points {E} new and the intermediate training data set D new(n×k) , in the way of matrix splicing, generate the training data set X, and the calculation formula of the corresponding training data set is X = {E new , D new}.

[0098] Step 5, according to the training data set X, calculate the hybrid intelligent model corresponding to the minimum value of the objective function, where the hybrid intelligent model is used for rapid evaluation of the mechanical properties of the automotive engine hood.

[0099] Specifically, as Figure 4 shown, through the intelligent fusion algorithm, fuse the model eigenvalue and the data set for eigenvalue fusion, realize the "intelligent adaptation" selection and retention of data and eigenvalues, and generate a new hybrid intelligent model.

[0100] In the fusion process, convert the training data set X in terms of expression, and successively extract the elements at the corresponding positions in the set of termination center points {E} new and the intermediate training data set D new(n×k) to form an array form, and convert the training data set X to:

[0101] where, x i′ =(x i′1 , x i′2 ,..., x i′d ) is the data in the set of termination center points {E} new , and y i is the data in the intermediate training data set D new(n×k) .

[0102] In the calculation process, let f(x)=ω T x i′ +b, y i′ =ω T x i′ +ε i′ , where,

[0103]

[0104] In the formula, i' is the row number, r is the total number of rows, d is the column number of the elements in the i'-th row, x i′ is the element in the i'-th row of the set of termination center points {E} new , y i′ is the element in the i'-th row of the data in the intermediate training data set D new(n×k) , ω is a preset constant, and ε i′ is the error value between the true value and the predicted value when using the elements in the i'-th row for the i'-th prediction, and σ is the standard deviation of the normal distribution.

[0105] Therefore, it can be calculated that:

[0106]

[0107] Let the objective function be:

[0108]

[0109] By making the objective function reach the minimum value, the model can be obtained:

[0110]

[0111] In the formula, is the model output, that is, the predicted value of the input data (training data set) after the establishment of this hybrid intelligent model, and z is the simulation value of the CAE software corresponding to the input data.

[0112] Therefore, the calculation formula of the hybrid intelligent model in this embodiment is:

[0113]

[0114] Using the above hybrid intelligent model to predict and evaluate the performance of the newly generated training data set X, the predicted performance of the first-order mode and the predicted performance of the torsional stiffness can be obtained respectively, as shown in Table 4 and Table 5 respectively.

[0115] Table 4

[0116] Prediction model <![CDATA[R 2 > MAE RMSE Fusion model 0.9969 0.63871 1.32185

[0117] Table 5

[0118] Prediction model <![CDATA[R 2 > MAE RMSE Fusion model 0.9877 0.00783 0.00989

[0119] As can be seen from Table 4 and Table 5, for the hybrid intelligent model constructed in this example, the algorithm performance index R 2 value, that is, the coefficient of determination of the model, is 0.9969 and 0.9877 respectively, and the mean absolute error MAE is 0.63871 and 0.00783 respectively, indicating that the prediction accuracy of this hybrid intelligent model is relatively high, and the root mean square error RMSE is 1.32185 and 0.00989 respectively, indicating that the abnormal amplitude of the outliers in the mechanical properties of the automotive engine hood predicted by this hybrid intelligent model is not large, indicating that the stability of the model prediction is good.

[0120] Moreover, through measurement, the time required for the hybrid intelligent model to run a set of data is only 0.04 seconds. While using the traditional CAE simulation method, in the same simulation environment, running the same set of data to achieve a similar simulation effect would take about 20 minutes. Therefore, in this embodiment, the hybrid intelligent model can significantly improve the design efficiency of engineers while ensuring the simulation accuracy.

[0121] The technical solution of the present application has been described in detail above with reference to the accompanying drawings. The present application proposes a method for rapidly evaluating the mechanical properties of an automotive engine hood based on machine learning. The method includes: Step 1, based on the original data set and multiple initial algorithm models, performing multiple trainings and tests, and extracting the initial feature vectors of the multiple initial algorithm models and the corresponding prediction data sets. Among them, the initial feature vectors at least include the first-order modal feature vectors and the torsional stiffness feature vectors; Step 2, calculating and generating an intermediate training data set according to the initial feature vectors and the prediction data sets; Step 3, calculating the termination center point set corresponding to the initial center point set when the preset condition function converges according to the initial center point set of the original data set and the preset condition function; Step 4, generating a training data set according to the termination center point set and the intermediate training data set; Step 5, calculating the hybrid intelligent model corresponding to the minimum value of the objective function value according to the training data set, where the hybrid intelligent model is used for rapidly evaluating the mechanical properties of the automotive engine hood. Through the technical solution in the present application, the problems of the long R & D cycle of the engine hood and the low prediction accuracy and stability are solved.

[0122] The steps in the present application can be adjusted, combined, and deleted according to actual needs.

[0123] The units in the device of the present application can be combined, divided, and deleted according to actual needs.

[0124] Although the present application has been disclosed in detail with reference to the accompanying drawings, it should be understood that these descriptions are merely exemplary and are not intended to limit the application of the present application. The protection scope of the present application is defined by the appended claims and may include various modifications, improvements, and equivalent solutions made to the invention without departing from the protection scope and spirit of the present application.

Claims

1. A rapid evaluation method for the mechanical properties of an automotive engine hood based on machine learning ， It is characterized in that This method includes: Step 1, based on the original data set and multiple initial algorithm models, conduct multiple trainings and tests, and extract the initial feature vectors of the multiple initial algorithm models and the corresponding prediction data sets. The initial feature vectors at least include the first-order modal feature vectors and the torsional stiffness feature vectors; Step 2, calculate and generate an intermediate training data set according to the initial feature vectors and the prediction data sets; Step 3, according to the initial center point set of the original data set and the preset condition function, calculate the termination center point set corresponding to the initial center point set when the preset condition function converges; Step 4, generate a training data set according to the termination center point set and the intermediate training data set; Step 5, calculate the hybrid intelligent model corresponding to the minimum value of the objective function value according to the training data set, where the hybrid intelligent model is used for rapid evaluation of the mechanical properties of the automotive engine hood; The calculation formula of the hybrid intelligent model is: Where X is the training data set, is the predicted value of the hybrid intelligent model, J(ω) is the objective function, ω is a preset constant, i′ is the row number, r is the total number of rows, x i′ is the set of termination center points {E} new the element in the i′-th row, y i′ is the element in the i′-th row of the data in the intermediate training data set D new(n×k) and z is the CAE software simulation value corresponding to the input data.

2. The rapid evaluation method for the mechanical properties of an automotive engine hood based on machine learning according to claim 1, It is characterized in that In step 2, the calculation and generation of the intermediate training data set according to the initial feature vectors and the prediction data sets specifically includes: Step 21, calculate the prediction mean of the prediction data set corresponding to each training and test and the maximum value in the prediction mean; Step 22, calculate and generate the intermediate training data set based on the initial feature vectors, the maximum value in the initial feature vectors, the prediction data sets, and the maximum value in the prediction mean.

3. The rapid evaluation method for the mechanical properties of an automotive engine hood based on machine learning according to claim 2, It is characterized in that In step 2, the intermediate training dataset D new(n×k) The corresponding calculation formula is: S M = Max{R ij} where D new(n×k) is the intermediate training data set, S M is the maximum value in the initial feature vector {R ij}, is the predicted mean of the prediction data set {Pred ij}, V M is the maximum value of the predicted mean in the prediction data set , and R ij is the element in the i-th row and j-th column of the feature vector {R ij}.

4. The rapid evaluation method for the mechanical properties of an automotive engine hood based on machine learning according to claim 3, It is characterized in that In step 3, it specifically includes: Step 31, use a clustering algorithm to determine the initial center point set of the original data set; Step 32, determine the category to which each training data in the original data set belongs according to the initial center point set; Step 33, according to the initial center point set and the category to which it belongs, use an iterative method to calculate the termination center point set corresponding to the initial center point set when the preset condition function converges.

5. The rapid evaluation method for the mechanical properties of an automotive engine hood based on machine learning according to claim 4, It is characterized in that In the said step 33, the termination center point set {E} new consists of multiple center points μ e The termination center point set {E} new The calculation formula for the center point μ e is as follows: where m is the number of the training data x m , m = 1, 2, …, M, c m is the category it belongs to, J(c, μ) is the preset conditional function, is the center point position of the category where the m-th data is located.

6. The rapid evaluation method for the mechanical properties of an automotive engine hood based on machine learning according to any one of claims 1 to 5, It is characterized in that Before step 5, it further includes: Convert the training data set X in terms of its expression, and sequentially extract the set of termination center points {E} new and the intermediate training data set D new(n×k) Extract the elements at the corresponding positions in the above to form an array, and convert the training data set X to:

Images